LLM Reference

Xiaomi MiMo-V2-Flash

xiaomi-mimo-v2-flash

Researched 17d ago

Last refreshed 2026-05-04. Next refresh: weekly.

RAGAgentsLong contextJSON / Tool use

Xiaomi MiMo-V2-Flash has model metadata, but missing tracked provider pricing keeps it from being a default production pick.

Decision context: RAG task fit, 0 tracked provider routes, and research from 2026-05-04.

Use it for

  • Teams evaluating rag, agents, and long context
  • Workloads that can use a 262K context window

Do not use it for

  • Cost-sensitive launches that need sourced token pricing
  • Vision or document-understanding workloads
  • Teams that need a tracked hosted API route today

Cheapest output

-

No tracked output price

Provider routes

0

No provider route in seed

Quality / dollar

Unknown

No task benchmark coverage yet

Freshness

2026-05-04

Researched 17d ago

fresh

Top use-case fit

RAG

Included by capability and metadata signals in the decision map.

Agents

Included by capability and metadata signals in the decision map.

Long context

Included by capability and metadata signals in the decision map.

Provider price ladder

No tracked provider token pricing is available for this model yet.

Benchmark peer barsfor RAG

No task-mapped benchmark peers are available for this model yet.

Migration checks

No linked migration route is available for this model yet.

About

MiMo-V2-Flash is Xiaomi's efficient open-source Mixture-of-Experts model, announced December 17, 2025 at Xiaomi's Human-Car-Home Ecosystem Partner Conference. It has 309B total parameters with 15B active, uses hybrid attention that interleaves Sliding Window Attention and Global Attention, and extends native 32K context to 256K. Multi-Token Prediction enables about 2.6x speculative decoding speedup. The model was distributed with weights on Hugging Face and ranked highly on SWE-Bench Verified and multilingual benchmarks at research time.

Xiaomi MiMo-V2-Flash has a 256K-token context window.

Capabilities

ReasoningFunction Calling

Rankings

Specifications

FamilyMiMo V2
Released2025-12-17
Parameters309B
Context262K
Architecturemoe
Knowledge cutoff2024-12
LicenseMIT
Trainingpretrained

Created by

Consumer electronics and AI research.

Beijing, China
Founded 2010
Website